Dual-center validation of a multi-sequence MRI radiomics nomogram for noninvasive assessment of renal fibrosis in chronic kidney disease

BACKGROUND: Accurate, noninvasive assessment of renal fibrosis (RF) in chronic kidney disease (CKD) remains challenging. This study aimed to develop and externally validate a dual-center multi-sequence MRI radiomics nomogram integrating imaging and clinical parameters for evaluating RF severity. METHODS: This retrospective dual-center study included 164 patients with CKD who underwent multi-sequence MRI and renal biopsy, divided into a training set (n = 128) and an external test set (n = 36). Radiomics features were extracted from intravoxel incoherent motion (IVIM) and blood oxygenation level-dependent (BOLD) MRI. Feature selection involved inter-observer correlation coefficient, Mann-Whitney U test, Pearson correlation coefficients, and least absolute shrinkage and selection operator regression. Three radiomics models (Rad_IVIM, Rad_BOLD, Rad_IVIM+BOLD) and one clinic model were developed to distinguish mild from moderate-to-severe RF. A nomogram was constructed by integrating the Rad_IVIM+BOLD score with significant clinical variables. Model performance was evaluated using area under the curve (AUC), DeLong test, decision curve analysis (DCA), and calibration curves. RESULTS: The Rad_IVIM+BOLD model achieved AUCs of 0.898 (95% confidence intervals [CI]: 0.845-0.951) in the training set and 0.749 (95% CI: 0.584-0.914) in the test sets, showing a favorable performance trend compared with the single-sequence radiomics models. The nomogram demonstrated good discrimination with AUCs of 0.922 (95% CI: 0.876-0.969) and 0.861 (95% CI: 0.728-0.993) in the training and test sets, respectively. The DeLong test showed its superiority over the clinic and radiomics models. DCA and calibration curves further confirmed its high predictive value and robustness. CONCLUSION: A nomogram combining multi-sequence MRI radiomics and clinical features provides a noninvasive tool for assessing RF severity in CKD, supporting quantitative imaging-guided clinical decision-making and disease follow-up.